The recent paper by Cerutti et al (1997) compared serum thrombopoietin (TPO) levels in patients with essential thrombocythaemia (ET) and reactive thrombocytosis (RT). Although serum TPO levels were significantly elevated in both patient groups compared to normal controls, the results suggest that the assay cannot be utilized to distinguish between ET and RT. It would be interesting to determine whether these results are reproducible in plasma, as platelet release of TPO may occur during activation and clotting.Human TPO plays a critical role in regulating megakaryocytopoiesis and platelet production (Kaushansky, 1995), hence regulating the number of reticulated platelets (young platelets), which can be detected by flow cytometric analysis of platelet RNA. Evidence is accumulating suggesting that the number of reticulated platelets is not only a reflection of platelet production but can be used to differentiate ET from RT (Harrison et al, 1998). We have recently utilized a modified method to study reticulated platelets (Robinson et al, 1998) in patients with untreated ET (n ¼ 10), RT (n ¼ 10) and normal controls (n ¼ 11). The percentage of reticulated platelets was significantly elevated in ET when compared to RT or controls (P < 0·05) (Fig 1A), in agreement with previous studies. However, although the absolute numbers (Fig 1B) of reticulated platelets were identical in ET and RT (P ¼ 0·29), they were both significantly elevated when compared to controls (P < 0·05). These results suggest that the overall platelet production and turnover is elevated in ET but not in RT. This could be a reflection of the elevated TPO levels in ET observed by Cerutti et al (1997) and Taraha et al (1996). The reduced expression of the c-mpl receptor copy number and hence reduced TPO clearance as recently described (Horikawa et al, 1997), could explain these observations in ET. The elevated TPO levels in RT, however, do not result in a significant effect on platelet production as reflected in the percentage of reticulated platelets. The increased absolute count in RT is merely a reflection of the relationship between platelet count and the percentage of reticulated platelets which should theoretically be linear when platelet production is normal.This data provides support that the pathogenesis of the elevated platelet count in RT is different from ET, but is unlikely to be solely the result of increased TPO production as suggested by Cerutti et al (1997), as this would result in an elevated percentage of reticulated platelets. In summary, determination of platelet mRNA may prove to be a useful positive diagnostic tool in the differentiation of ET from RT. Larger studies correlating serum or plasma TPO levels and reticulated platelets in thrombocytosis will clarify the issue further.